Btukfyl

I want to create a bar chart from the following data, using the column difficulty to select rows from kappa.

I hope to avoid having to type out the contents of the csv file in a series of add plot statements

Can one use the meta variable as follows?

begin{axis}[ybar]



 begin{tikzpicture}

begin{axis}[ybar,ymajorgrids, tick align=inside,

    major grid style={draw=white},axis x line*=bottom,

    axis y line*=right,symbolic x coords={

       Human-Human,Human-Tak,Human-Gold Standard, Tak-Gold Standard},]

    addplot table[x=type, y=kappa, meta=difficulty=0]{combined-kappas.csv};

    addplot table[x=type, y=kappa, meta=difficulty=1]{combined-kappas.csv};

    addplot table[x=type, y=kappa, meta=difficulty=2]{combined-kappas.csv};

end{axis}

CSV Data:

r1,r2,type,kappa,difficulty

Alex,Tak,Human-Human,0.9878019029031471,0

Alex,Inferred,Human-Tak,0.6199032480995161,0

Alex,Intended,Human-Gold Standard,1,0

Tak,Inferred,Human-Tak,0.6081143384047948,0

Tak,Intended,Human-Gold Standard,0.9878019029031471,0

Intended,Inferred,Tak-Gold Standard,0.6199032480995161,0

Alex,Tak,Human-Human, 0.791208791208791,1

Alex,Inferred,Human-Tak,0.306901088386556,1

Alex,Intended,Human-Gold Standard,0.844389844389844,1

Tak,Inferred,Human-Tak,0.245305729417429,1

Tak,Intended,Human-Gold Standard,0.757327238504975,1

Intended,Inferred,Tak-Gold Standard,0.337421837421837,1

Alex,Tak,Human-Human,0.547137761168677,2

Alex,Inferred,Human-Tak,0.137205030380104,2

Alex,Intended,Human-Gold Standard,0.262046204620462,2

Tak,Inferred,Human-Tak,0.137288621381485,2

Tak,Intended,Human-Gold Standard,0.280682000965096,2

Intended,Inferred,Tak-Gold Standard,0.097688926697859,2

Intended output:

To answer to your comment below the question:

I think the linked answer in my comment is what you are searching for. But there is one remaining problem that is, that you have multiple entries for one bar sometimes which results in multiple bars plotted. In your image there is some kind of error bar given to the bars.

The error bars can also be given to bars, but of course for that you need to modify your CSV file accordingly.

(Please note that this answer is just to prove that this in principle is what you are searching for and thus the resulting image does not look very nice.)

% used PGFPlots v1.16

begin{filecontents*}{combined-kappas.csv}

r1,r2,type,kappa,difficulty

Alex,Tak,Human-Human,0.9878019029031471,0

Alex,Inferred,Human-Tak,0.6199032480995161,0

Alex,Intended,Human-Gold Standard,1,0

Tak,Inferred,Human-Tak,0.6081143384047948,0

Tak,Intended,Human-Gold Standard,0.9878019029031471,0

Intended,Inferred,Tak-Gold Standard,0.6199032480995161,0

Alex,Tak,Human-Human, 0.791208791208791,1

Alex,Inferred,Human-Tak,0.306901088386556,1

Alex,Intended,Human-Gold Standard,0.844389844389844,1

Tak,Inferred,Human-Tak,0.245305729417429,1

Tak,Intended,Human-Gold Standard,0.757327238504975,1

Intended,Inferred,Tak-Gold Standard,0.337421837421837,1

Alex,Tak,Human-Human,0.547137761168677,2

Alex,Inferred,Human-Tak,0.137205030380104,2

Alex,Intended,Human-Gold Standard,0.262046204620462,2

Tak,Inferred,Human-Tak,0.137288621381485,2

Tak,Intended,Human-Gold Standard,0.280682000965096,2

Intended,Inferred,Tak-Gold Standard,0.097688926697859,2

end{filecontents*}

documentclass[border=5pt]{standalone}

usepackage{pgfplots}

    % borrowed styles from Jake's answer at

    % <http://tex.stackexchange.com/a/301991/95441>

    pgfplotsset{

%        % filter symbolic values

%        discard if not symbolic/.style 2 args={

%            % suppress LOG messages about the filtered points

%            filter discard warning=false,

%            x filter/.append code={

%                edeftempa{thisrow{#1}}

%                edeftempb{#2}

%                ifxtempatempb

%                else

%                    defpgfmathresult{NaN}

%                fi

%            },

%        },

        % filter numerical values

        % (not needed here, but just for completeness)

        discard if not/.style 2 args={

            % suppress LOG messages about the filtered points

            filter discard warning=false,

            x filter/.append code={

                ifdimthisrow{#1} pt=#2pt

                else

                    defpgfmathresult{NaN}

                fi

            },

        },

    }

begin{document}

    begin{tikzpicture}

        begin{axis}[

            ybar,

            symbolic x coords={

                Human-Human,

                Human-Tak,

                Human-Gold Standard,

                Tak-Gold Standard

            },

            enlarge x limits=0.2,

        ]

            foreach i in {0,1,2} {

                addplot+ [

                    discard if not={difficulty}{i},

                ] table [

                    col sep=comma,

                    x=type,

                    y=kappa,

                ] {combined-kappas.csv};

                    addlegendentryexpanded{i}

            }

        end{axis}

    end{tikzpicture}

end{document}